HAL Id: hal-01408896https://hal.archives-ouvertes.fr/hal-01408896

Submitted on 4 Oct 2018

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License

D-Serine and Glycine Differentially ControlNeurotransmission during Visual Cortex Critical PeriodClaire Nicole-Jeanne Meunier, Glenn Dallérac, Nicolas Le Roux, Silvia Sacchi,Grégoire Levasseur, Muriel Amar, Loredano Pollegioni, Jean-Pierre Mothet,

Philippe Fossier

To cite this version:Claire Nicole-Jeanne Meunier, Glenn Dallérac, Nicolas Le Roux, Silvia Sacchi, Grégoire Levasseur,et al.. D-Serine and Glycine Differentially Control Neurotransmission during Visual Cortex Crit-ical Period. PLoS ONE, Public Library of Science, 2016, 11 (3), pp.e0151233. �10.1371/jour-nal.pone.0151233�. �hal-01408896�

RESEARCH ARTICLE

D-Serine and Glycine Differentially ControlNeurotransmission during Visual CortexCritical PeriodClaire N. J. Meunier1¤‡, Glenn Dallérac2¤‡, Nicolas Le Roux1‡, Silvia Sacchi3,4,5,Grégoire Levasseur2, Muriel Amar1, Loredano Pollegioni3,4,5, Jean-Pierre Mothet2*,Philippe Fossier1*

1 Institut de Neuroscience Paris-Saclay (NeuroPSI), UMR 9197 CNRS-Université Paris-Sud, Bât 446, F-91405, Orsay cedex, France, 2 Aix-Marseille University, CRN2MUMR7286 CNRS, 51 Bd Pierre Dramard,13344, Marseille, France, 3 Dipartimento di Biotecnologie e Scienze della Vita, Università degli Studidell’Insubria, via J.H. Dunant 3, Varese, Italy, 4 “The Protein Factory”, Centro Interuniversitario diBiotecnologie Proteiche, Politecnico di Milano, ICRM-CNR, Milano, Italy, 5 Università degli Studidell’Insubria, via Mancinelli 7, Milano, Italy

¤ Current address: Collège de France, CIRB, CNRS UMR 7241, INSERMU1050, F-75005, Paris, France‡ These authors are co-first authors.* philippe.fossier@u-psud.fr (PF); jean-pierre.mothet@univ-amu.fr (JPM)

AbstractN-methyl-D-aspartate receptors (NMDARs) play a central role in synaptic plasticity. Their

activation requires the binding of both glutamate and D-serine or glycine as co-agonist. The

prevalence of either co-agonist on NMDA-receptor function differs between brain regions

and remains undetermined in the visual cortex (VC) at the critical period of postnatal devel-

opment. Here, we therefore investigated the regulatory role that D-serine and/or glycine may

exert on NMDARs function and on synaptic plasticity in the rat VC layer 5 pyramidal neu-

rons of young rats. Using selective enzymatic depletion of D-serine or glycine, we demon-

strate that D-serine and not glycine is the endogenous co-agonist of synaptic NMDARs

required for the induction and expression of Long Term Potentiation (LTP) at both excitatory

and inhibitory synapses. Glycine on the other hand is not involved in synaptic efficacy perse but regulates excitatory and inhibitory neurotransmission by activating strychnine-sensi-

tive glycine receptors, then producing a shunting inhibition that controls neuronal gain and

results in a depression of synaptic inputs at the somatic level after dendritic integration. In

conclusion, we describe for the first time that in the VC both D-serine and glycine differen-

tially regulate somatic depolarization through the activation of distinct synaptic and extrasy-

naptic receptors.

IntroductionN-Methyl-D-aspartate receptors (NMDARs) are central for structural and functional synapticplasticity as well as cognitive functions [1]. Activation of such receptor requires the binding ofboth glutamate and a co-agonist [2]. Although glycine was initially identified as the main co-agonist [3,4], subsequent investigations revealed that D-serine, synthesized by serine racemase

PLOSONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 1 / 20

OPEN ACCESS

Citation: Meunier CNJ, Dallérac G, Le Roux N,Sacchi S, Levasseur G, Amar M, et al. (2016) D-Serine and Glycine Differentially ControlNeurotransmission during Visual Cortex CriticalPeriod. PLoS ONE 11(3): e0151233. doi:10.1371/journal.pone.0151233

Editor: Shao-Jun Tang, University of Texas MedicalBranch, UNITED STATES

Received: June 5, 2015

Accepted: February 25, 2016

Published: March 22, 2016

Copyright: © 2016 Meunier et al. This is an openaccess article distributed under the terms of theCreative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in anymedium, provided the original author and source arecredited.

Data Availability Statement: All relevant data arewithin the paper and its Supporting Information files.

Funding: This research was supported by grantsfrom the Centre National pour la RechercheScientifique (CNRS), SANOFI-AVENTIS and AgenceNationale de la Recherche (ANR). CNJM received astudentship from CNRS and SANOFI-AVENTIS RDExploratory Unit. JPM and GD were supported byAgence Nationale pour la Recherche (Grant numberANR-09-MNPS-022-01), CNRS, Université Aix-Marseille and Fondation pour la Recherche Médicale(to J-P.M.). The funders had no role in study design,

(SR) [5,6] and present in areas where NMDARs are prevalent [7], would be the preferred co-agonist for synaptic NMDARs [8]. Indeed, enzymatically or genetically induced depletion of D-serine reduces synaptic NMDARs currents and thereby alters synaptic plasticity in the hippo-campus [9–12], prefrontal cortex [13], and nucleus accumbens [14]. The role of D-serine atNMDARs is further illustrated by studies showing that synaptic and cognitive impairmentsduring aging is linked to a down-regulation of D-serine synthesis [15]. Detailed analysis of thecontribution of the two co-agonists in NMDARs regulation further reveals that in the CA1area of the mature hippocampus D-serine would preferentially act on synaptic NMDARs whilstglycine would modulate extrasynaptic NMDARs [12] although this segregation has beenshown to be developmentally regulated [16]. Alternatively, Li and colleagues (2013) proposethat the prevalence of D-serine or glycine at synaptic NMDARs in the lateral nucleus of theamygdala would rather be determined by synaptic activity [17] a scenario also reported for thehippocampus [16]. Despite major progress in the definition of D-serine and glycine functionsat excitatory synapses, the nature of the endogenous co-agonist in primary sensory areas likethe visual cortex remains to be defined notably during the critical period of enhanced plasticityenabling activity-dependent proper development and maturation of the visual system.

Abundant evidence points to the importance of NMDARs in patterning neuronal networksin the visual cortex [18,19]. Indeed, NMDARs-regulated neurotransmission has been suggestedto play an important role in ocular dominance (OD) plasticity in both juvenile and adultrodents [20]. Pharmacological blockade or genetic deletions of NMDARs prevent the OD shiftafter monocular deprivation (MD) during the critical period or in adult [21,22]. Two recentstudies have investigated the functional contribution of D-serine in the plasticity of the visualcortex. Yang and colleagues have shown that D-serine facilitates adult cortical NMDAR-depen-dent synaptic and OD plasticity in MDmice [23]. Furthermore, D-serine depletion by the enzy-matic scavenger D-amino acid oxidase causes NMDAR-dependent phase coupling of otherwisephase-independent gamma generating networks, causing hypersynchrony and a distortion ofvisual perception [24]. These latter observations suggest that D-serine may serve as an endoge-nous ligand for VC NMDARs and is necessary for proper visual processing. However, it is stillunknown how D-serine and glycine interplay modulates excitatory and inhibitory neuronalnetworks in the visual cortex.

Here, we use enzymatically-driven depletion of D-serine and glycine levels to ascertain theirfunctions in neurotransmission and synaptic plasticity in acute VC slices of P19-25 old rats.Using whole-cell patch clamp recordings of postsynaptic currents enabling to assess inhibitoryand excitatory synaptic conductances at layer 5 pyramidal neurons (L5PyNs), we demonstratethat selective loss of function of D-serine but not glycine reduces synaptic events and preventsinduction and expression of NMDAR-dependent inhibitory and excitatory LTP in the VC.Furthermore we show that, in contrast, glycine does not modulate synaptic plasticity per se butacts at the dendritic integration level, through the activation of strychnine-sensitive glycinereceptors (GlyRs), and thereby controls neuronal gain. The present study therefore shows that,in the VC, control of somatic depolarization depends on synaptic plasticity as such, whichrequires D-serine, as well as on the modulation of signal integration through the opening ofGlyRs.

Experimental Procedures

Slices preparation and electrophysiology recordingsExperiments were carried out between September 2010 and December 2014 on male Wistarrats in accordance with the European and Institutional guidelines for the care and use of labo-ratory animals (Council Directive 86/609/EEC and 2010/63/UE and its application in 2013 by

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 2 / 20

data collection and analysis, decision to publish, orpreparation of the manuscript.

Competing Interests: Part of the work of CNJM wasfunded by Sanofi-Aventis. However, this does notalter the authors’ adherence to PLOS ONE policieson sharing data and materials.

the French National Research Council. Rats were bred under standard conditions in our animalfacilitity licensed by the French veterinary service (INSERM U894-CPN; CNPS). Agreementnumber of the animal house facility is C 91-471-104. Our study includes exclusively in vitroexperiments with no in vivo work. The article 3 of the 2010/63/UE directive permits to euthan-atize animals by cervical dislocation to excise brain tissues for experiments without anyrequirement of a specific ethical committee agreement. P21 to P28 rats were killed by cervicaldislocation and their brains quickly removed. CNJM, who hold a license (number R-45GRE-TA-F1-10) delivered by the French veterinary service, sacrificed all animals used in this study.

Wistar male rats aged 21 to 28 days old were subject to the decapitation procedure and afterquick removal of the brain, one hemisphere was removed, attached to the stage of a tissue slicer(WPI NVSLM1, U.K) and immersed in ice-cold, oxygenated (i.e., bubbled with 95% O2/5%CO2) artificial cerebrospinal fluid (ACSF) containing (in mM): 126 NaCl, 26 NaHCO3, 10 Glu-cose, 2 CaCl2, 1.5 KCl, 1.5 MgSO4 and 1.25 KH2PO4 (pH 7.5, 310–320 mOsm). Parasagittalslices (250 μm) containing primary visual cortex were obtained and transferred to an holdingchamber filled with oxygenated ACSF and maintained at room temperature after an initial 1hincubation at 36°C. For experiments, slices were perfused (2–3 ml/min) at 31°C with oxygen-ated ACSF. Neurons were patched at 40x magnification using an upright microscope (ZeissAxioscop FS2+). Patch-clamp recording pipettes (4–5 MΩ) were filled with intracellular solu-tion (in mM): 140 K-gluconate, 10 HEPES, 4 ATP, 2 MgCl2, 0.4 GTP and 0.5 EGTA (adjustedto pH 7.3 with KOH, 280–290 mOsm.kg-1). Stable whole-cell voltage-clamp recordings wereobtained from layer 5 pyramidal neurons (L5PyNs, identified by the shape of their soma andmain apical dendrite and from their firing profile induced by 1s depolarizing steps rangingfrom 0 to 200 pA) with a Multiclamp 700A amplifier (Axon Instruments, USA). Data weresampled at 2 kHz using a Digidata 1322A acquisition board (Axon Instruments, USA). Voltagedata were corrected off-line for a measured liquid junction potential of -10 mV. After capaci-tance neutralization, bridge balancing was performed on-line under current clamp to make ini-tial estimations of the access resistance (Rs). The latter procedure was repeated before everyvoltage clamp recording. The membrane input resistance (Rm) and time constant (Ƭ0) weredetermined off line by fitting the mean voltage response to a short hyperpolarizing currentpulse applied at rest in current clamp mode. Only cells with a resting potential less than -60mV and with an access resistance lower than 25 MΩ were considered for analysis. Recordingswith more than 25% change in input resistance were also discarded.

Recording of the NMDA-Excitatory Postsynaptic Currents (EPSCs) inL5PyNsNMDA-EPSCs were recorded in response to electrical stimulation of layer 2/3 after blockadeof GABAA receptors with picrotoxin (100 μM) and AMPA receptors with NBQX (10 μM) at aholding potential of +40 mV using an intracellular solution containing (in mM): Cs-methylsul-fonate, 115; HEPES, 10; ATP, 4; CsCl, 20; GTP, 0.4; EGTA, 10 (adjusted to pH 7.4 with CsOH;281 mOsm.kg-1). The origin of the recorded current was confirmed by its complete blockadeafter application of the NMDA receptor antagonist CPP (1μM) (S1 Fig). At least 5 recordingswere stacked and averaged.

Stimulation protocolsFor Excitation–Inhibition (E-I) balance determination, electrical stimulations (10–100 μA, 0.2ms, 0.05 Hz) were applied in the layer 2/3 of visual cortex using 1 MΩ impedance bipolar tung-sten electrodes. Electrodes were positioned in the vicinity of L5PyNs apical dendrite in order torecruit feedforward monosynaptic excitation and inhibition as well as disynaptic inhibition

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 3 / 20

[25]. The stimulation intensity was adjusted in current-clamp [25,26] and set to 2–3 times theamplitude of the stimulation necessary to induce a detectable response in current clamp, whichhas been shown to generate linear subthreshold postsynaptic responses resulting fom coactiva-tion of excitatory and inhibitory circuits. Under voltage-clamp, 4 to 8 trials were repeated for 4to 7 holding potentials (depending on the cell resting membrane potential). For LTP induction,theta-burst stimulation (TBS) was applied in layer 2/3 only. TBS consisted of 3 trains of 13bursts applied at a frequency of 5 Hz, each burst containing four pulses at 100 Hz. Inter-traininterval was 10 s. The recording protocol was set as follows: baseline recording of the currentresponses enabling determination of the different conductances (gT, gE, gI) was performed 10min after settling of the whole cell patch clamp; then the drug was applied and composite cur-rent recordings were resumed pre-TBS as well as 15, 30, 45, and 60 min post-TBS. Pre-TBSrecordings were performed while drugs were washing. LTP protocols were applied immediatelyafter the pre-TBS recording.

Determination of the E-I balanceData were analysed off-line with Acquis1TM and ElphyTM (Biologic UNIC–CNRS, Gif-sur-Yvette, France). The E-I balance determination is based on the continuous measurement ofconductance dynamics during the full-time course of the stimulus-evoked synaptic response.Briefly, we performed post-hoc decomposition of postsynaptic current waveforms in excitatoryand inhibitory conductances together with continuous estimation of the apparent reversalpotential of the composite responses. This allows a somatic measurement of the E-I balance inL5PyNs after dendritic integration of incoming excitation and inhibition [27].

In order to extract the excitatory and inhibitory conductance changes from the evoked syn-aptic currents, the neuron is considered as the point-conductance model of a single-compart-ment cell, described by the following general membrane equation:

CmdVmðtÞ

dt¼ gleakðVmðtÞ � EleakÞ � gEðtÞðVmðtÞ � EexcÞ � gIðtÞðVmðtÞ � EinhÞ þ Iinj

where Cm denotes the membrane capacitance, Iinj the injected current, gleak the leak conduc-tance and Eleak the leak reversal potential. gE(t) and gI(t) are the excitatory and inhibitory con-ductances, with respective reversal potentials Eexc and Einh.

Evoked synaptic currents were measured and averaged for several (4 to 8) holding poten-tials. IV curves were then calculated at all time points of the response. In IV curves for everypossible delay (t), the value of holding potential (Vh) was corrected (Vhc) from the ohmicdrop due to leakage current through the access resistance (Vhc(t) = Vh(t)–I(t) x Rs). An aver-age estimate of the input conductance waveform of the cell was calculated from the best linearfit (mean least square criterion) of the IV curve for each delay (t) following the stimulationonset. Only cells showing a Pearson correlation coefficient for the IV linear regression higherthan 0.95 between –90 and –40 mV were considered for calculation of the conductance changein the recorded pyramidal neuron, using the slope of the regression line. The synapticallyevoked global conductance term (gT(t)) was then measured by subtracting the resting conduc-tance observed in the absence of stimulation (on a time window of 100 ms before electricalstimulation) from the input total conductance. The synaptic reversal potential of the synapticconductance (Esyn(t)) was taken as the voltage of the intersection between the IV curve duringthe synaptic response and the IV curve at rest. Assuming that the evoked somatic conductancechange reflects the composite synaptic input reaching the soma, Esyn(t) characterizes the stim-ulation-locked dynamics of the balance between excitation and inhibition. The global synapticconductance (gT(t)) was further decomposed into two conductance components (gE(t) and gI

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 4 / 20

(t)) corresponding to the activation of excitatory and inhibitory synapses respectively, eachassociated with known and fixed reversal potentials. Indeed, we showed [25] that the IV curvein the presence of excitatory transmission blockers (CNQX, D-AP5) is linear between -80 to+10 mV with a reversal potential equal to -80 mV. In the presence of bicuculline, blockinginhibitory inputs on the L5PyNs, the IV curve for excitation is also linear between -80 to +10mV with a reversal potential equal to 0 mV. Furthermore, D-AP5 had no impact on the evokedcurrent responses recorded at 0.05Hz suggesting that NMDA receptors are not recruited.

Accordingly, the reversal potentials used for the decomposition of the global synaptic con-ductance were set at 0 mV for excitatory (Eexc) and -80 mV for inhibitory conductance (Einh).In addition, these reversal potentials correspond to values typically found in other studies [28].Einh corresponds to the reversal potential of GABAA (and not an intermediate value betweenGABAA and GABAB) because in the presence of QX314 in the pipette no variation of the syn-aptic response was observed [27]. Under our experimental conditions, Esyn(t) took any inter-mediate values between Eexc (0 mV) and Einh (-80 mV) in such a way that the mathematicalconditions of the simplification used to calculate gI(t) and gE(t) were fulfilled. For each compo-nent, excitatory and inhibitory, we calculated the conductance change as the mean averagedover a time window of 200 ms. The contribution of each component was expressed by the ratioof its integral value (intgE or intgI) to that of global conductance change (intgT).

Like all somatic recordings, our recordings cannot make rigorous estimates of synapticevents in the distal dendrites, and estimated conductances are ratios of the overall excitatoryand inhibitory drive contained in the local network stimulated [29]. However, our measure-ments are relative changes in conductance magnitude which reflect the cumulative contribu-tions of excitation and inhibition arriving at proximal portions of the neuron. These relativeconductance changes at the somatic level define a narrow window over which input integrationand spike output can occur [30].

Determination of the time constants and electrotonic lengthMembrane time constants (Ƭ0) were determined by analyzing the time course of the membranevoltage deflection according to the method described by Rall [31]. This method consists in“peeling exponentials” by plotting the natural log of the response expressed as a percentage ofthe peak negative potential. In our recordings, the late portion of the charging phase (for timepoints between 5 and 15 ms) always fitted a linear regression of which the slope was equal to-1/Ƭ0. For each cell and hyperpolarizing intensity, the membrane time constant was thus calcu-lated using this equation. Rall’s method also allows determination of the equalizing time con-stant (Ƭ1), which represents the time required for the current to spread to the proximaldendrites. Ƭ1 is given by the second order exponential observed for time points earlier than5ms. This value was obtained by plotting the difference between the membrane time constantregression line from the points lying above this line, normalizing the y-intercept of this newline to 100% and reading Ƭ1 as the negative inverse of the slope. From two time constants, thesomatodendritic electrotonic length (L) was estimated using the relation: L = π (Ƭ0/ Ƭ 1)

-1/2.

Statistical analysisData reported are mean ± standard error of the mean (SEM) of n cells. Each individual cell wasrecorded in a single acute slice (3–4 slices/ animal per day) in order to avoid repeated exposureto the TBS protocol and/or drug application. When a LTP protocol was applied (TBS), post-TBS conductance integral values were normalized to pre-TBS conductance integrals (100%)and expressed as percentage of baseline. LTP was defined as a change>130% of baseline 1 hpost-TBS. Statistics were performed using the InVivoStat software (Mockett Media). Paired

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 5 / 20

samples for IntgT, IntgE, IntgI, between the control condition (before TBS) and given timesafter TBS (15, 30, 45 or 60 min) were analyzed using the paired student t-test.

Chemicals and enzymesAll chemicals used for electrophysiology experiments were obtained from Sigma-Aldrichexcept NBQX which were from Ascent Scientific (Bristol UK). Appropriate stock solutionswere made, stored at −20°C and diluted to the final concentration in aCSF.

The selective depletion of D-serine and glycine was obtained by using recombinantRgDAAO (EC 1.4.3.3) and recombinant BsGO (EC 1.4.3.19) which were overexpressed inEscherichia coli cells and purified as previously reported [32,33]. The final enzyme preparationsdisplayed the following activity: RgDAAO = 100 ± 15 U/mg protein on D-serine as substrate;BsGO = 0.9 ± 0.2 U/mg protein on glycine as substrate. These flavoenzymes specificallydegrade D-serine (RgDAAO) and glycine (BsGO), as demonstrated by the correspondingapparent kinetic efficiency (kcat/Km ratio) values: RgDAAO kcat/Km ratios were 3.0 and 0.058mM−1 s−1 [34], while those determined for BsGO were 0.00025 and 0.867 mM−1 s−1 on D-serineand glycine, respectively [33]. Enzymatic treatments were performed by incubating for at least45 min and then continuously perfusing the slices with ACSF containing RgDAAO (0.2 U/ml)or BsGO (0.1 U/ml). An inactive variant of RgDAAO (ΔRgDAAO) that does not dehydroge-nate D-amino acids due to the substitution of the pivotal active site residue Arg285 with Ala-nine [35] was used as control. Enzymes activity is not altered by the presence of picrotoxin(100 μM) and NBQX (10 μM) during EPSCs recordings [12,16].

ImmunohistochemistryRats were deeply anesthetized with pentobarbital (6%) and perfused transcardially with phos-phate buffer (PB) 0.1 M (pH 7.3) and paraformaldehyde (4%) supplemented with 0.25% glutar-aldehyde. The brain was post-fixed overnight in the same solution and finally cryoprotectedwith 30% sucrose in 0.1M PB. Brains were sliced (30 μm) using a vibratome and immunostain-ings were performed as previously described [13]. Immunostained sections were observed witha ZEISS LSM 710 confocal microscope (ZEN software) with 405 diode, 488 and 633 laser lines.All image acquisitions were achieved using 40X and 60X magnification and variable numericalzooms. Channels were acquired with sequential scanning with a pinhole aperture of 1 airy uniteach. Images were then analyzed using ImageJ software. Staining specificity was verified withnegative controls in which primary or secondary antibodies were omitted. Affinity-purifiedprimary antibodies were: mouse monoclonal anti serine racemase (Transduction Laboratories:1/1000), rabbit polyclonal anti-D-serine (GemacBio: 1/1000), mouse polyclonal anti-GFAP(Sigma: 1/2000, clone G-A-5), mouse monoclonal anti-glyR (mAb4a, Synaptic Systems 1/100).Secondary antibodies were obtained from Invitrogen: Alexa 488-Goat anti-mouse (1/1000),Alexa 633-Goat anti-rabbit (1/1000) and from Life Technologies: FITC conjugated mouse anti-body (1/400).

Results

d-Serine is the endogenous agonist of synaptic NMDARs in the VCWe first assessed the putative contribution of D-serine in the VC by immunostainings for D-ser-ine. We found D-serine and the D-serine producing enzyme serine racemase to be present inall layers including layer 5 of the VC (Fig 1A), at P22-P25 suggesting that D-serine may alreadycontribute to synapse and neuronal networks functions during the first weeks of postnataldevelopment. We then assessed the functions of D-serine in driving synaptic activity by

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 6 / 20

recording postsynaptic NMDA currents (NMDA-EPSCs) in L5PyNs. Bath application of theselective D-serine scavenger D-amino acid oxidase (RgDAAO, 0.2 U/ml) [13,16] decreasedNMDA-EPSCs by 29.3 ± 5.1% (Fig 1B and 1D; n = 5 cells, 5 slices, 2 animals P<0.001) to asimilar order of magnitude as the co-agonist site blocker 7-Cl-KYN (S1 Fig), while the inactivevariant of RgDAAO (ΔRgDAAO) had no effect (Fig 1D). Conversely, bath application of D-ser-ine (100 μM) increased NMDA-EPSCs by 46.3 ± 7.5% (Fig 1C & 1D; n = 5 cells, 5 slices, 2 ani-mals, P<0.001) thus showing that the co-agonist site of NMDARs is not saturated in the VC,at least during the time of critical period for plasticity.

Fig 1. D-serine modulates synaptic NMDARs current at VC L5PyNsA: Immunofluorescence for D-serine, and GFAP revealed that D-serine is expressedacross all layers of P22-25 (5 slices, 3 animals) VC. The D-serine producing enzyme serine racemase (SR) was also found to be expressed and co-localizedwith the astroglial marker GFAP. Images are Z-stack of 10 serial confocal images with a thickness of 1μm. Scale bars: left, 200 μm; right, 50μm B-D:Applications of the D-serine degrading enzyme RgDAAO (0.2 U/ml) reduces synaptically evoked NMDA-EPSCS (n = 5 cells, 5 slices, 2 animals) (B) while itsinactive form ΔRgDAAO has no effect (n = 5 cells, 5 slices, 2 animals) (D). Scale bars: 100pA, 500ms. Conversely D-serine (100 μM) significantly potentiatesNMDA-EPSCs (n = 5 cells, 5 slices, 2 animals) (C) Scale bars: 200pA, 500ms. ***p<0.001.

doi:10.1371/journal.pone.0151233.g001

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 7 / 20

Because RgDAAO only partially reduced the amplitude of NMDA-EPSCs (Fig 1B & 1D),we then addressed whether glycine in addition to D-serine could also be a co-agonist for synap-tic NMDARs in VC of young rats. Depletion of endogenous glycine level using the recombi-nant Bacillus subtilis glycine oxidase (BsGO, 0.1U/ml) did not affect NMDA-EPSCs amplitude(Fig 2A & 2F; n = 5 cells, 5 slices, 2 animals, P>0.05). Because such lack of effect could be dueto failure of BsGO to reduce glycine levels [13,16,36] we performed control experiments withALX5407 (2 μM) to increase endogenous glycine levels by inhibiting glycine transporters type1 (GlyT1) and then applied BsGO (Fig 2B). ALX5407 decreased rather than increasedNMDA-EPSCs amplitude by 20.3 ± 3.1% (Fig 2B & 2F; n = 4 cells, 4 slices, 2 animals, P<0.01),an effect indeed reversed by BsGO (Fig 2B & 2F) then showing that the latter was effective indepleting endogenous glycine. Strikingly, bath application of glycine (100μM) significantlydecreased NMDA-EPSCs amplitude by 21.9 ± 4.2% (Fig 2C & 2F; n = 5 cells, 5 slices, 2 animals,P<0.001). These observations indicate that D-serine rather than glycine is the co-agonist ofsynaptic NMDARs in the VC of young rats under normal conditions.

Insofar as BsGO did not alter NMDA-EPSCs, we hypothesized that the down neuromodula-tion exerted by glycine on excitatory neurotransmission may occur through activation ofstrychnine-sensitive glycine receptors (GlyRs), as seen in the hippocampus [37]. To test thishypothesis, we blocked GlyRs with bath application of 10 μM strychnine and showed that suchtreatment was sufficient to prevent downregulation of NMDA-EPSCs by 100 μM glycine (Fig2D & 2F; n = 5 cells, 5 slices, 2 animals). Besides, we verified the presence of these receptors byperforming immunostainings for GlyRs [38]. Fig 2E shows that GlyRs are indeed abundant inapical dendrites and soma of pyramidal neurons indicating their possible role in integratinginformation at L5PyNs of VC.

Altogether, these data support that in L5PyNs of VC, D-serine regulates synaptic NMDARsefficacy by acting as the co-agonist of NMDARs, while glycine acts downstream through acti-vation of dendritic and somatic GlyRs.

d-Serine enables visual cortex LTPHaving established that D-serine is the endogenous ligand of NMDARs in the young visual cor-tex, we next examined its contribution to long-term potentiation (LTP) by determining thetotal synaptic conductance change (referred as IntgT) and by analyzing both excitatory (eLTP)and inhibitory (iLTP) components of the response. We have previously shown that ThetaBurst Stimulation (TBS) of layer 2–3 of the VC resulted in similar levels of eLTP and iLTPrecorded at the soma of L5PyNs and that this form of LTP depends on NMDARs functions[25]. Accordingly, in the present study, TBS protocol induced comparable LTP (Fig 3A & 3B;IntgT: 148.7 ± 3.0% of baseline, Ps<0.001) and resulted in similar levels of eLTP and iLTP (Fig3B & 3C; IntgE: 146.7 ± 4.0% of baseline, IntgI: 150.1 ± 3.0% of baseline, Ps<0.001, n = 15cells, 15 slices, 8 animals), hence not modifying the E-I balance. Most interestingly, bath appli-cation of the NMDARs co-agonist site blocker 7Cl-KYN (50 μM) [39] during and after TBSprotocol not only prevented LTP but depressed synaptic responses after TBS administration(IntgT: 68.5 ± 6.3% of baseline, P<0.001; n = 18 cells, 18 slices, 9 animals) at both excitatoryand inhibitory level (Fig 3B & 3F; IntgE: 76.8 ± 6.2% of baseline, P<0.01; IntgI: 67.7 ± 6.8% ofbaseline, P<0.001). TBS induced LTP is therefore, in the VC, NMDA-receptor dependent, asexpected from our previous work [40].

Accordingly, depletion of endogenous D-serine with RgDAAO (0.2 U/ml) prevented induc-tion of LTP (Fig 3B & 3D, IntgT:105.4 ± 8.3% of baseline; IntgE:111.8 ± 10.8% of baseline;IntgI: 106.4 ± 10.4% of baseline, n = 8 cells, 8 slices, 4 animals, Ps>0.05) while application of itsinactive form ΔRgDAAO throughout the experiment showed no effect (Fig 3B; IntgT:

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 8 / 20

Fig 2. Glycine is not the endogenous co-agonist of L5PyNs VC NMDARsA: Application of the glycine degrading enzyme BsGO (0.1 U/ml) has no effect onNMDA-EPSCs (n = 5 cells, 5 slices, 2 animals), indicating that glycine is not the endogenous co-agonist of synaptic L5PyrNs VC NMDARs. Scale bars:100pA, 500ms.B: Further, enhancing endogenous glycine levels with the glycine transporter blocker ALX5407 (2 μM) decreased the NMDARs response, aneffect blocked by BsGO (n = 4 cells, 4 slices, 2 animals). Scale bars: 25pA, 250ms.C: A similar result is obtained by exogenous application of glycine(100μM) (n = 5 cells, 5 slices, 2 animals). Scale bars: 50pA, 500ms. D: Such downregulation of NMDA-EPSCs by glycine is remarkably blocked by theglycinergic receptors (GyRs) antagonist strychnine (10μM) (n = 5 cells, 5 slices, 2 animals). Scale bars: 50pA, 500ms. E: Immunofluorescence for GlyRs

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 9 / 20

154.8 ± 12.9% of baseline; IntgE: 144.3 ± 10.5% of baseline; IntgI: 162.1 ± 16.2% of baseline;n = 15 cells, 15 slices, 8 animals, Ps<0.001), thus confirming the specificity of RgDAAOblockade.

Interestingly, in contrast with 7Cl-KYN, RgDAAO failed to engender TBS-induced synapticdepression. Since this apparent discrepancy could reflect incomplete depletion of endogenousD-serine by RgDAAO we further tested the contribution of D-serine in LTP using phenazineEthosulfate (Et-Phen), an inhibitor of serine racemase, throughout the experiment [41–43].Pharmacological inhibition of SR not only prevented the TBS-induced eLTP and iLTP asobserved with RgDAAO (Fig 3B & 3E) but also induced depression of the synaptic responses,as observed with 7Cl-KYN (Fig 3B & 3E; IntgT: 75.9 ± 7.4% of baseline; IntgE: 77.3 ± 7.6% ofbaseline; IntgI: 75.3 ± 8.6% of baseline; n = 17 cells, 17 slices, 8 animals, Ps<0.01). Altogether,the dataset therefore indicates that D-serine is required for both eLTP and iLTP and that itsabsence unmasks a depression of evoked synaptic responses recorded at the soma of L5PyNs.

Glycine downregulates current spread in VC L5PyNsGiven the lack of effect of BsGO on NMDA-EPSCs and their downregulation in the presenceof glycine, we then sought to ascertain to which extent endogenous glycine may or not contrib-ute to the modulation of L5PyNs synaptic current during higher regime of activity. To this end,glycine depletion was achieved again by application of BsGO (0.1 U/ml) throughout the experi-ment. Under these conditions, LTP was found to be unchanged (Fig 4B & 4C; IntgT:136.4 ± 6.4% of baseline; IntgE: 139.8 ± 9.7% of baseline; IntgI: 134.2% ± 9.0% of baseline;n = 19 cells, 19 slices, 9 animals, Ps<0.01) indicating that unlike D-serine, reduced level of gly-cine does not limit LTP induction and expression in the rat VC. We also tested whether, con-versely, increase in endogenous glycine could affect LTP. To this end we applied the selectiveblocker of GlyT1 ALX5407 (2μM) half an hour before TBS and throughout the experiment[36] in order to build up glycine concentration in the cortical slice. Under ALX5407, TBS failedto induce LTP (Fig 4B & 4D; IntgT: 101.2 ± 8.0% of baseline; IntgE: 110.7 ± 9.3% of baseline;IntgI: 104.2% ± 7.4% of baseline, Ps>0.05 n = 13 cells, 13 slices, 6 animals). We further con-firmed that this effect was attributable to the action of glycine as adding BsGO (0.1 U/ml) inthe presence of ALX5407 (2μM) throughout the experiment showed no effect on VC LTP (Fig4B; IntgT: 133.2 ± 4.8% of baseline, IntgE: 135.4 ± 6.2% of baseline and IntgI: 131.1 ± 4.0% ofbaseline respectively; Ps> 0.01, n = 13 cells, 13 slices, 6 animals). This result was identical tothe result obtained in the presence of BsGO alone. Since BsGO was effective in depleting gly-cine, we conclude that raising endogenous glycine concentration prevents somatic recording ofLTP in L5PyNs.

To further characterize the dose-dependent effect of glycine on TBS induced plasticity, wemeasured the effect of three concentrations of glycine on excitatory and inhibitory plasticity.Application of 1 μM glycine was sufficient to annihilate LTP (Fig 5A & 5B; IntgT: 120.2 ± 8.1%of baseline; IntgE: 114.4 ± 8.8% of baseline; IntgI: 127.3 ± 7.5% of baseline, n = 10 cells, 10slices, 5 animals, P>0.05). At higher concentrations, exogenous glycine (10 μM) resulted in aweak depression following TBS (Fig 5A & 5C; IntgT: 84.5 ± 6.2% of baseline; IntgE:81.6 ± 7.6% of baseline; IntgI: 85.5 ± 7.9% of baseline, n = 10 cells, 10 slices, 5 animals, Ps>0.05), an effect significantly amplified at 100 μM (Fig 5A & 5D; IntgT: 66.8 ± 5.7%; IntgE:74.4 ± 7.2% of baseline; IntgI: 62.2 ± 6 .4% of baseline; n = 12 cells, 12 slices, 6 animals,Ps<0.001). Insofar as we found that glycine is able to downregulate NMDARs currents through

revealed that, in the VC, they are mainly expressed in L5PyRNs notably at the somatic and dendritic level. Scale bar: 50μm, inset: 30μm. F: Altogether, theseresults indicate that glycine downregulates NMDA-EPSCs through activation of GlyRs ** p<0.01, *** p<0.001.

doi:10.1371/journal.pone.0151233.g002

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 10 / 20

Fig 3. D-serine is required for VC long-term potentiationA:Upper panel shows representative composite current responses of L5PyN for the range ofimposed potentials before and during LTP. Scale bars: 300pA, 50ms. Medium panels displays the corresponding total conductance change gT before andduring LTP. Lower panels show decomposition of gT into excitatory (gE, black) and inhibitory (gI, grey) conductances (n = 15 cells, 15 slices, 8 animals).Scale bars: 4nS, 50ms.B: Changes in the gT integral (IntgT) calculated every 15 min, up to 1 h post-TBS, show that LTP is abolished by blocking the co-agonist binding site of NMDARs with 7-Cl-KYN, removing D-serine through the D-serine degrading enzyme RgDAAO (n = 18 cells, 18 slices, 9 animals) orpreventing D-serine production via blockade of the D-serine producing enzyme serine racemase with Et-Phen (n = 17 cells, 17 slices, 8 animals). Thisindicates that D-serine is required for VC L5PyNs LTP.C-F: Excitatory (black) and inhibitory (grey) conductances were found to be equally affected by TBSapplication, regardless of the treatment, indicating that the E-I balance is unaltered by D-serine and LTP. *p<0.05, **p<0.01, ***p<0.001 compared to pre-TBS.

doi:10.1371/journal.pone.0151233.g003

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 11 / 20

activation of GlyRs, we analyzed contribution of the latter by applying TBS in the presence ofglycine (100 μM) and the GlyR blocker strychnine (10 μM). This resulted in an LTP compara-ble to that obtained in control conditions (Fig 5A & 5E; IntgT: 138.9 ± 6.7% of baseline; IntgE:137.3 ± 7.9% of baseline; IntgI: 140.7% ± 6.7% of baseline; n = 12 cells, 12 slices, 6 animals, Ps>0.01). Furthermore, applying TBS whilst blocking the NMDARs co-agonist site with 7-Cl-KYNin the presence of glycine (100μM) still depressed synaptic responses (S2 Fig; IntgT:73.9 ± 11.4% of baseline; IntgE: 77.1 ± 10.3% of baseline; IntgI: 62.5 ± 12.3% of baseline; n = 4,Ps<0.05) indicating that NMDARs are not involved in these downregulations. Finally, in thepresence of 7-Cl-KYN and strychnine TBS induced no change (S2 Fig; IntgT: 101.2 ± 10.9% ofbaseline; IntgE: 103.5 ± 12.7% of baseline; IntgI: 91.5 ± 10.2% of baseline; n = 4; Ps>0.05) in

Fig 4. Increasing endogenous glycine level prevents the induction of LTP. A: Upper panel shows representative composite current responses of L5PyNfor the range of imposed potentials before and during LTP. Scale bars: 150pA, 50ms. Medium panels displays the corresponding total conductance changegT before and during LTP. Lower panels show decomposition of gT into excitatory (gE, black) and inhibitory (gI, grey) conductances. Scale bars: 4nS, 50ms.B: Changes in IntgT up to 1 h post-TBS show that LTP is abolished when endogenous glycine levels are increased by blocking the glycine transporter withALX (2μM) (n = 13 cells, 13 slices, 6 animals). The glycine degrading enzyme BsGO has no effect on LTP (n = 19 cells, 19 slices, 9 animals), and preventsthe effect of ALX (n = 13 cells, 13 slices, 6 animals), thus confirming that the latter is attributable to endogenous glycine rise.C-D: Excitatory (black) andinhibitory (grey) conductances were found to be equally affected by TBS application, regardless of the treatment, indicating that the E-I balance is unalteredby glycine and LTP. *p<0.05, **p<0.01, ***p<0.001 compared to pre-TBS.

doi:10.1371/journal.pone.0151233.g004

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 12 / 20

Fig 5. Dose effects of various concentrations of glycine on LTPA: At the concentration of 1μM glycine the initial potentiation induced by TBS does not last 1h, thus indicating that at such low concentration is enough to prevent LTP (n = 10 cells, 10 slices, 5 animals). At 10μM glycine not only prevents allpotentiation but also slightly decreases conductances recorded at the soma (n = 10 cells, 10 slices, 5 animals), although this reduction is not significant. At100μM glycine blocks LTP and induces a significant depression up to 1 h post-TBS (n = 12 cells, 12 slices, 6 animals).B-E: Excitatory (black) and inhibitory

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 13 / 20

current magnitude, thus implying that GlyRs activation underlies the reduction of recordedcurrents. We therefore conclude that increased levels of glycine during high regime of activitysuch as TBS activate GlyRs and eventually leads to a depression of the conductances recordedafter TBS at the somatic level. Given the extrasynaptic dendritic and somatic localisation ofGlyRs, we hypothesized that their activation results in a shunt of synaptic inputs.

To assess such possibility, we calculated the electrotonic length of L5PyN and used it as anindex of current spread efficacy according to the method described by Rall [31]. Remarkably,the electrotonic length was found to be significantly higher in the presence of 100 μM glycine(Fig 6B; Control: 0.52 ± 0.03, n = 15; Glycine: 0.73 ± 0.03, n = 16 cells, 16 slices, 8 animals;P<0.01), an effect that corresponds to an increased attenuation of the distal currents reading atthe somatic level [44,45]. We conclude that activation of dendritic GlyRs results in a filteringeffect. In all, these data indicate that, in the VC, activation of GlyRs by elevated endogenousglycine levels during high regime of activity filters synaptic inputs conveyed through the den-dritic tree, thereby masking LTP induced at distal synapses and resulting in LTD-like changesin current readings at the somatic level.

DiscussionThe present study provides evidence that D-serine controls NMDAR-dependent LTP in VCL5PyNs whilst glycine influence neurotransmission at a different level, by activating extrasy-naptic GlyRs distributed along the apical dendrite. Activation of the GlyRs when the concen-tration of glycine increases indeed results in a shunting inhibition of afferent inputs which thusdisplay a depression (a LTD-like effect) instead of an LTP at the soma after dendriticintegration.

Whilst former investigations have supported that glycine is the co-agonist of NMDARs[3,4,46,47], recent work have shown that reducing D-serine levels impairs NMDAR-mediatedprocesses in several structures, including the hippocampus, prefrontal cortex, nucleus accum-bens or amygdala [9,10,12–14,17,48], suggesting that D-serine is likely to be the co-agonist forsynaptic NMDAR prevailing in many brain areas, even though glycine remains engaged in themodulation these receptors [11, 16, 17]. Using D-serine and/or glycine, several reports indicatethat the co-agonist site of synaptic NMDARs is not saturated in the VC of cats and rats, duringand after the critical period of plasticity [49–52]. However, thus far, identity of the prevalentco-agonist remained undetermined in this structure. Our data show that, at VC L5PyNs, reduc-ing D-serine function using a variety of pharmacological treatments prevents the induction ofLTP whilst depleting glycine has no effect, thus demonstrating in the VC that D-serine and notglycine is the dominant endogenous ligand of synaptic NMDARs, as seen in other brain areas[9,12,13,16,53]. It however remains possible that modulation of extrasynaptic NMDARs is dif-ferent and involves glycine.

Both D-serine and glycine have been reported to be present in the micromolar range in vari-ous cortical areas [7,54], although their concentrations may differ between the in vivo vs exvivo situation. Insofar as our results indicate that glycine does not undertake the main role ofNMDA receptor co-agonist, we further investigated its role in this brain region. The presenceof GlyRs along the dendritic tree of L5PyNs suggests a functional role at this level. Interestinglyhowever, removing glycine from the preparation using BsGO throughout the experimentshowed no effect on NMDA receptor function. Thus, considering the inhibitory effect weobserved on NMDA currents and the increase in electrotonic length engendered by glycine

(grey) conductances were found to be equally affected by TBS application, regardless of the treatment, confirming that the E-I balance is unaltered by glycineand LTP. *p<0.05, **p<0.01, ***p<0.001 compared to pre-TBS.

doi:10.1371/journal.pone.0151233.g005

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 14 / 20

application, we propose that by opening GlyRs, rise of glycine that occurs primarily duringhigh regime of activity would allow for a potent filtering of synaptic inputs conveyed throughthe neuron. Although membrane depolarizations may be large and sharp locally at the synapse,the resulting changes in somatic potential are much slower and smaller after dendritic propaga-tion [55]. Insofar as opening of chloride permeability at resting potential mainly acts as a shuntsince the equilibrium potential for Cl- is -80mV [25], activation of non-synaptic dendriticGlyRs should indeed accentuate current attenuation upon propagation. Such interpretation issupported by our observation that downregulation of NMDA currents and LTP by glycine isfully prevented by the GlyRs blocker strychnine. A similar effect has also been observed in thehippocampus where GlyRs act as a tonic shunt, thereby depressing excitatory neurotransmis-sion and inducing LTD [37,56,57]. The present study thus shows that the situation is mostlikely similar in the VC, where EPSCs arising at the synaptic level are filtered and depressedalong the dendritic tree of L5PyNs upon opening of GlyRs, this results in a LTD-like plasticityat the soma. Such selective activation of GlyRs and not NMDARs in the VC is most likelyattributable to the differential sites of action of these receptors. It is indeed readily conceivablethat glycine remains confined to extrasynaptic sites, where GlyRs are expressed, as glycinetransporters can prevent its access to synaptic confinements, as found in the hippocampus[12]. In the cortex and hippocampus, GlyRs have been shown to be primarily expressed atsomatic and dendritic extrasynaptic sites [58,59]. These anatomical observations are also cor-roborated by a large body of physiological investigations reporting that synaptic currents areabolished by antagonists of glutamate and GABA receptors in the postnatal cortex, includingour work in the VC [60,61], and thereby suggesting that cortical GlyRs are not synaptically

Fig 6. Glycine reduces VC L5PyNs dendritic current spreadA:Membrane time constants Ƭ0 and Ƭ1. were calculated by plotting the natural log of theresponse expressed as percentage of the peak negative potential (% ΔEmax) to “peel” the first order exponential for time points lying between 5 and 15ms.Ƭ0 could then be read as the slope negative inverse of the regression line. The second order exponential for time points earlier than 5ms was peeled byplotting the difference between the Ƭ0 regression line from the points lying above this line and normalizing the y-intercept of the Ƭ1 regression line to 100%. Ƭ1could then be read as the slope negative inverse of the normalized Ƭ1 regression line. B:Calculation of the electrotonic length (L) using the equation L = π(Ƭ0/ Ƭ 1)

-1/2 showed that L is significantly increased in the presence of 100μM glycine, indicating that the dentritic current spread attenuation is higher in thelatter case. **p<0.01,***p<0.001.

doi:10.1371/journal.pone.0151233.g006

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 15 / 20

activated. Such data support the early demonstration by Flint and colleagues that GlyRs areactivated non-synaptically [62]; consistent with this finding, a recent report found that applica-tion of GlyR antagonist strychnine dampens membrane currents, without affecting spontane-ous synaptic events [63]. In all, our findings thus represent a new framework integrating thecomplex functions of D-serine and glycine in the modulation of synapse activity and thedynamics of VC neuronal networks.

The brain uses a variety of strategies to process and extract information upon the largerange of sensory inputs it receives. Several investigations have sought the optimal strategies thebrain may employ to tune the best signal-to-noise ratio obtainable, such as, for example, gaincontrol [64]. Yet, in primary sensory cortices, the strategies to scale integration of the inputstrength and extract relevant information from noise remain poorly understood [65,66]. Oneinteresting model postulates that the VC would select pertinent information by a noise-filteringaction, dampening responses to irrelevant visual noise [67]. Remarkably, a recent study pro-vides fMRI data supporting the latter hypothesis [68]. By demonstrating that activation ofGlyRs along the L5PyNs dendrite enables efficient shunt of synaptic inputs while maintainingthe E-I balance intact, the present study shed light on a mechanism that likely participates inthe selective wave attenuation of sensory inputs, known to be necessary for the adaptation ofinputs intensity to visual processing [69].

Supporting InformationS1 Fig. NMDAR currents are antagonized by CPP and 7-Cl-KYN. A: Administration of theselective NMDARs antagonist CPP (1μM) expectedly abolished the recorded current, thus con-firming their nature (n = 4). B: Bath application of the co-agonist site blocker 7-Cl-KYNdecreased NMDA-EPSCs to the same extent as the selective D-serine scavenger D-amino acidoxidase thus suggesting that D-serine is the endogenous co-agonist of NMDARs in the visualcortex (n = 3).(TIF)

S2 Fig. TBS induced depression depends on GlyRs and not NMDARs. A: The depressionobserved after TBS administration in the presence of glycine 100μM is not affected by theNMDAR co-agonist binding site blocker 7-Cl-KYN (50μM) indicating that putative regulationof NMDAR by glycine does not play a role in this process (n = 4). B: Instead, GlyRs underliesuch downregulation as the GlyRs blocker strychnine (10μM) abolishes the TBS induceddepression (n = 4). �p<0.05, ��p<0.01, ���p<0.001.(TIF)

AcknowledgmentsThis research was supported by grants from CNRS. CNJM received a studentship from CNRSand SANOFI-AVENTIS RD Exploratory Unit. JPM and GDwere supported by Agence Natio-nale pour la Recherche (Grant number ANR-09-MNPS-022-01), CNRS, Université Aix-Mar-seille and Fondation pour la Recherche Médicale (to J-P.M.). The authors are grateful to Dr CyrilMonier who wrote and provided support with the analysis software. The funders had no role instudy design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author ContributionsConceived and designed the experiments: PF JPM LP GDMA. Performed the experiments:CNJM NL GL GD SS. Analyzed the data: CNJM GD PF SS. Wrote the paper: GD CNJM PFJPM.

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 16 / 20

References1. Collingridge GL, Volianskis A, Bannister N, France G, Hanna L, Mercier M, et al. (2013) The NMDA

receptor as a target for cognitive enhancement. Neuropharmacology 64: 13–26. doi: 10.1016/j.neuropharm.2012.06.051 PMID: 22796429

2. Paoletti P, Bellone C, Zhou Q (2013) NMDA receptor subunit diversity: impact on receptor properties,synaptic plasticity and disease. Nat Rev Neurosci 14: 383–400. doi: 10.1038/nrn3504 PMID:23686171

3. Johnson JW, Ascher P (1987) Glycine potentiates the NMDA response in cultured mouse brain neu-rons. Nature 325: 529–531. doi: 10.1038/325529a0 PMID: 2433595

4. Kleckner NW, Dingledine R (1988) Requirement for glycine in activation of NMDA-receptors expressedin Xenopus oocytes. Science 241: 835–837. PMID: 2841759

5. Wolosker H, Sheth KN, Takahashi M, Mothet JP, Brady RO, Ferris CD, et al. (1999) Purification of ser-ine racemase: biosynthesis of the neuromodulator D-serine. Proc Natl Acad Sci U S A 96: 721–725.PMID: 9892700

6. Wolosker H, Blackshaw S, Snyder SH (1999) Serine racemase: a glial enzyme synthesizing D-serineto regulate glutamate-N-methyl-D-aspartate neurotransmission. Proc Natl Acad Sci U S A 96: 13409–13414. PMID: 10557334

7. Schell MJ, Molliver ME, Snyder SH (1995) D-serine, an endogenous synaptic modulator: localization toastrocytes and glutamate-stimulated release. Proc Natl Acad Sci U S A 92: 3948–3952. PMID:7732010

8. Martineau M, Baux G, Mothet J-P (2006) D-serine signalling in the brain: friend and foe. Trends Neu-rosci 29: 481–491. doi: 10.1016/j.tins.2006.06.008 PMID: 16806506

9. Mothet JP, Parent AT, Wolosker H, Brady RO, Linden DJ, Ferris CD, et al. (2000) D-serine is an endog-enous ligand for the glycine site of the N-methyl-D-aspartate receptor. Proc Natl Acad Sci U S A 97:4926–4931. PMID: 10781100

10. Basu AC, Tsai GE, Ma C-L, Ehmsen JT, Mustafa AK, Han L, et al. (2009) Targeted disruption of serineracemase affects glutamatergic neurotransmission and behavior. Mol Psychiatry 14: 719–727. doi: 10.1038/mp.2008.130 PMID: 19065142

11. Rosenberg D, Artoul S, Segal AC, Kolodney G, Radzishevsky I, Dikopoltsev E, et al. (2013) NeuronalD-Serine and Glycine Release Via the Asc-1 Transporter Regulates NMDAReceptor-Dependent Syn-aptic Activity. J Neurosci 33: 3533–3544. doi: 10.1523/JNEUROSCI.3836-12.2013 PMID: 23426681

12. Papouin T, Ladépêche L, Ruel J, Sacchi S, Labasque M, Hanini M, et al. (2012) Synaptic and extrasy-naptic NMDA receptors are gated by different endogenous coagonists. Cell 150: 633–646. doi: 10.1016/j.cell.2012.06.029 PMID: 22863013

13. Fossat P, Turpin FR, Sacchi S, Dulong J, Shi T, Rivet JM, et al. (2011) Glial D-Serine Gates NMDAReceptors at Excitatory Synapses in Prefrontal Cortex. Cereb Cortex: 1–12. doi: 10.1093/cercor/bhr130

14. Curcio L, Podda M V, Leone L, Piacentini R, Mastrodonato A, Cappelletti P, et al. (2013) Reduced D-serine levels in the nucleus accumbens of cocaine-treated rats hinder the induction of NMDA receptor-dependent synaptic plasticity. Brain 136: 1216–1230. doi: 10.1093/brain/awt036 PMID: 23518710

15. Turpin FR, Potier B, Dulong JR, Sinet P-M, Alliot J, Oliet SH, et al. (2011) Reduced serine racemaseexpression contributes to age-related deficits in hippocampal cognitive function. Neurobiol Aging 32:1495–1504. doi: 10.1016/j.neurobiolaging.2009.09.001 PMID: 19800712

16. Le Bail M, Martineau M, Sacchi S, Yatsenko N, Radzishevsky I, Conrod S, et al. (2014) Identity of theNMDA receptor coagonist is synapse specific and developmentally regulated in the hippocampus. ProcNatl Acad Sci U S A. doi: 10.1073/pnas.1416668112

17. Li Y, Sacchi S, Pollegioni L, Basu AC, Coyle JT, Bolshakov VY, et al. (2013) Identity of endogenousNMDAR glycine site agonist in amygdala is determined by synaptic activity level. Nat Commun 4:1760. doi: 10.1038/ncomms2779 PMID: 23612301

18. Quinlan EM, Philpot BD, Huganir RL, Bear MF (1999) Rapid, experience-dependent expression of syn-aptic NMDA receptors in visual cortex in vivo. Nat Neurosci 2: 352–357. doi: 10.1038/7263 PMID:10204542

19. Yashiro K, Philpot BD (2008) Regulation of NMDA receptor subunit expression and its implications forLTD, LTP, and metaplasticity. Neuropharmacology 55: 1081–1094. doi: 10.1016/j.neuropharm.2008.07.046 PMID: 18755202

20. Cho KKA, Khibnik L, Philpot BD, Bear MF (2009) The ratio of NR2A/B NMDA receptor subunits deter-mines the qualities of ocular dominance plasticity in visual cortex. Proc Natl Acad Sci U S A 106: 5377–5382. doi: 10.1073/pnas.0808104106 PMID: 19276107

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 17 / 20

21. Cao Z, Liu L, Lickey M, Graves A, Pham T, Gordon B,(2007) Virally mediated knock-down of NR2 sub-units ipsilateral to the deprived eye blocks ocular dominance plasticity. Exp brain Res 177: 64–77. doi:10.1007/s00221-006-0647-8 PMID: 16944113

22. Krahe TE, Medina AE (2010) Activation of NMDA receptors is necessary for the recovery of cortical bin-ocularity. J Neurophysiol 103: 2700–2706. doi: 10.1152/jn.00442.2009 PMID: 20457852

23. Yang K, XiongW, Yang G, Kojic L, Taghibiglou C, Wang YT, et al. (2011) The regulatory role of long-term depression in juvenile and adult mouse ocular dominance plasticity. Sci Rep 1: 203. doi: 10.1038/srep00203 PMID: 22355718

24. Anver H, Ward PD, Magony A, Vreugdenhil M (2011) NMDA receptor hypofunction phase couples inde-pendent γ-oscillations in the rat visual cortex. Neuropsychopharmacology 36: 519–528. doi: 10.1038/npp.2010.183 PMID: 20962769

25. Le Roux N, Amar M, Baux G, Fossier P (2006) Homeostatic control of the excitation-inhibition balancein cortical layer 5 pyramidal neurons. Eur J Neurosci 24: 3507–3518. doi: 10.1111/j.1460-9568.2006.05203.x PMID: 17229099

26. Moreau AW, Amar M, Le Roux N, Morel N, Fossier P (2010) Serotoninergic fine-tuning of the excita-tion-inhibition balance in rat visual cortical networks. Cereb Cortex 20: 456–467. doi: 10.1093/cercor/bhp114 PMID: 19520765

27. Monier C, Fournier J, Frégnac Y (2008) In vitro and in vivo measures of evoked excitatory and inhibitoryconductance dynamics in sensory cortices. J Neurosci Methods 169: 323–365. doi: 10.1016/j.jneumeth.2007.11.008 PMID: 18215425

28. Wehr M, Zador A (2003) Balanced inhibition underlies tuning and sharpens spike timing in auditory cor-tex. Nature 426: 442–446. doi: 10.1038/nature02116 PMID: 14647382

29. Haider B, Duque A (2006) Neocortical network activity in vivo is generated through a dynamic balanceof excitation and inhibition. J . . . 26: 4535–4545. doi: 10.1523/JNEUROSCI.5297-05.2006

30. Higley M, Contreras D (2006) Balanced excitation and inhibition determine spike timing during fre-quency adaptation. J Neurosci 26: 448–457. doi: 10.1523/JNEUROSCI.3506-05.2006 PMID:16407542

31. Rall W (1969) Time constants and electrotonic length of membrane cylinders and neurons. Biophys J9: 1483–1508. PMID: 5352228

32. Fantinato S, Pollegioni L PM (2001) Engineering, expression and purification of a His-tagged chimericD-amino acid oxidase from Rhodotorula gracilis. EnzymMicrob Technol 29: 407–412.

33. Job V, Marcone GL, Pilone MS, Pollegioni L (2002) Glycine oxidase from Bacillus subtilis. Characteri-zation of a new flavoprotein. J Biol Chem 277: 6985–6993. doi: 10.1074/jbc.M111095200 PMID:11744710

34. Pollegioni L, Piubelli L, Sacchi S, Pilone MS, Molla G (2007) Physiological functions of D-amino acidoxidases: from yeast to humans. Cell Mol Life Sci 64: 1373–1394. doi: 10.1007/s00018-007-6558-4PMID: 17396222

35. Molla G, Porrini D, Job V, Motteran L, Vegezzi C, Campaner S, et al. (2000) Role of arginine 285 in theactive site of Rhodotorula gracilis D-amino acid oxidase. A site-directed mutagenesis study. J BiolChem 275: 24715–24721. doi: 10.1074/jbc.M908193199 PMID: 10821840

36. Atkinson BN, Bell SC, De Vivo M, Kowalski LR, Lechner SM, Ognyanov VI, et al. (2001) ALX 5407: apotent, selective inhibitor of the hGlyT1 glycine transporter. Mol Pharmacol 60: 1414–1420. PMID:11723250

37. Chen R-Q, Wang S-H, YaoW, Wang J-J, Ji F, Yan J-Z, et al. (2011) Role of glycine receptors in gly-cine-induced LTD in hippocampal CA1 pyramidal neurons. Neuropsychopharmacology 36: 1948–1958. doi: 10.1038/npp.2011.86 PMID: 21593734

38. Danglot L, Rostaing P, Triller A, Bessis A (2004) Morphologically identified glycinergic synapses in thehippocampus. Mol Cell Neurosci 27: 394–403. doi: 10.1016/j.mcn.2004.05.007 PMID: 15555918

39. Krasteniakov N V, Martina M, Bergeron R (2005) Role of the glycine site of the N-methyl-D-aspartatereceptor in synaptic plasticity induced by pairing. Eur J Neurosci 21: 2782–2792. doi: 10.1111/j.1460-9568.2005.04099.x PMID: 15926925

40. Le Roux N, Amar M, Moreau A, Fossier P (2007) Involvement of NR2A- or NR2B-containing N-methyl-D-aspartate receptors in the potentiation of cortical layer 5 pyramidal neurone inputs depends on thedevelopmental stage. Eur J Neurosci 26: 289–301. doi: 10.1111/j.1460-9568.2007.05671.x PMID:17650107

41. Kim PM, Aizawa H, Kim PS, Huang AS, Wickramasinghe SR, Kashani AH, et al. (2005) Serine race-mase: activation by glutamate neurotransmission via glutamate receptor interacting protein and media-tion of neuronal migration. Proc Natl Acad Sci U S A 102: 2105–2110. doi: 10.1073/pnas.0409723102PMID: 15684087

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 18 / 20

42. Stevens ER, Gustafson EC, Sullivan SJ, Esguerra M, Miller RF (2010) Light-evoked NMDA receptor-mediated currents are reduced by blocking D-serine synthesis in the salamander retina. Neuroreport21: 239–244. doi: 10.1097/WNR.0b013e32833313b7 PMID: 20101193

43. Hagiwara H, Iyo M, Hashimoto K (2013) Neonatal disruption of serine racemase causes schizophrenia-like behavioral abnormalities in adulthood: clinical rescue by d-serine. PLoS One 8: e62438. doi: 10.1371/journal.pone.0062438 PMID: 23630632

44. Tuckwell HC, Rospars JP, Vermeulen a, Lánsky P (1996) Time-dependent solutions for a cable modelof an olfactory receptor neuron. J Theor Biol 181: 25–31. doi: 10.1006/jtbi.1996.0111 PMID: 8796188

45. Spruston N, Jaffe DB, Johnston D (1994) Dendritic attenuation of synaptic potentials and currents: therole of passive membrane properties. Trends Neurosci 17: 161–166. PMID: 7517596

46. Bergeron R, Meyer TM, Coyle JT, Greene RW (1998) Modulation of N-methyl-D-aspartate receptorfunction by glycine transport. Proc Natl Acad Sci U S A 95: 15730–15734. PMID: 9861038

47. Tsai G, Ralph-Williams RJ, Martina M, Bergeron R, Berger-Sweeney J, Dunham KS, et al. (2004) Geneknockout of glycine transporter 1: characterization of the behavioral phenotype. Proc Natl Acad Sci U SA 101: 8485–8490. doi: 10.1073/pnas.0402662101 PMID: 15159536

48. Wake K, Yamazaki H, Hanzawa S, Konno R, Sakio H, Niwa A, et al. (2001) Exaggerated responses tochronic nociceptive stimuli and enhancement of N-methyl-D-aspartate receptor-mediated synaptictransmission in mutant mice lacking D-amino-acid oxidase. Neurosci Lett 297: 25–28. PMID:11114476

49. Li Y-H, Han T-Z (2008) Glycinemodulates synaptic NR2A- and NR2B-containing NMDA receptor-medi-ated responses in the rat visual cortex. Brain Res 1190: 49–55. doi: 10.1016/j.brainres.2007.11.006PMID: 18048007

50. Li Y-H, Han T-Z (2007) Glycine binding sites of presynaptic NMDA receptors may tonically regulate glu-tamate release in the rat visual cortex. J Neurophysiol 97: 817–823. doi: 10.1152/jn.00980.2006 PMID:17093111

51. Ito K, Hicks TP (2001) Effect of the glycine modulatory site of the N-methyl-D-aspartate receptor on syn-aptic responses in kitten visual cortex. Neurosci Lett 303: 95–98. PMID: 11311501

52. Czepita D, Daw NW, Reid SN (1996) Glycine at the NMDA receptor in cat visual cortex: saturation andchanges with age. J Neurophysiol 75: 311–317. PMID: 8822559

53. Panatier A, Theodosis DT, Mothet J-P, Touquet B, Pollegioni L, Poulain DA, et al. (2006) Glia-derivedD-serine controls NMDA receptor activity and synaptic memory. Cell 125: 775–784. doi: 10.1016/j.cell.2006.02.051 PMID: 16713567

54. Hashimoto A, Okas T, Nishikawas T (1995) Extracellular concentration of endogenous free S-serine inthe rat brain as revealed by in vivo microdialysis. Neuroscience 66: 635–643. PMID: 7644027

55. Magee JC (2000) Dendritic integration of excitatory synaptic input. Nat Rev Neurosci 1: 181–190. doi:10.1038/35044552 PMID: 11257906

56. SongW, Chattipakorn SC, McMahon LL (2006) Glycine-gated chloride channels depress synaptictransmission in rat hippocampus. J Neurophysiol 95: 2366–2379. doi: 10.1152/jn.00386.2005 PMID:16381810

57. Keck T, Lillis KP, Zhou Y-D, White JA (2008) Frequency-dependent glycinergic inhibition modulatesplasticity in hippocampus. J Neurosci 28: 7359–7369. doi: 10.1523/JNEUROSCI.5618-07.2008 PMID:18632940

58. Becker CM, Betz H, Schröder H (1993) Expression of inhibitory glycine receptors in postnatal rat cere-bral cortex. Brain Res 606: 220–226. PMID: 8387859

59. Aroeira RI, Ribeiro JA, Sebastião AM, Valente CA (2011) Age-related changes of glycine receptor atthe rat hippocampus: from the embryo to the adult. J Neurochem 118: 339–353. doi: 10.1111/j.1471-4159.2011.07197.x PMID: 21272003

60. Le Roux N, Amar M, Moreau A, Baux G, Fossier P (2008) Impaired GABAergic transmission disruptsnormal homeostatic plasticity in rat cortical networks. Eur J Neurosci 27: 3244–3256. doi: 10.1111/j.1460-9568.2008.06288.x PMID: 18598264

61. Lucas-Meunier E, Monier C, Amar M, Baux G, Frégnac Y, Fossier P (2009) Involvement of nicotinicand muscarinic receptors in the endogenous cholinergic modulation of the balance between excitationand inhibition in the young rat visual cortex. Cereb Cortex 19: 2411–2427. doi: 10.1093/cercor/bhn258PMID: 19176636

62. Flint AC, Liu X, Kriegstein AR (1998) Nonsynaptic Glycine Receptor Activation during Early NeocorticalDevelopment. Neuron 20: 43–53. doi: 10.1016/S0896-6273(00)80433-X PMID: 9459441

63. Salling MC, Harrison NL (2014) Strychnine-sensitive glycine receptors on pyramidal neurons in layersII/III of the mouse prefrontal cortex are tonically activated. J Neurophysiol 112: 1169–1178. doi: 10.1152/jn.00714.2013 PMID: 24872538

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 19 / 20

64. Schwartz O SE (2001) Natural signal statistics and sensory gain control. Nat Neurosci 4: 819–825.PMID: 11477428

65. Levitt JB LJ (1997) Contrast dependence of contextual effects in primate visual cortex. PLoS One 387:73–76.

66. Kapadia MK, Westheimer G GC (1999) Dynamics of spatial summation in primary visual cortex of alertmonkeys. Proc Natl Acad Sci USA 96: 12073–12078. PMID: 10518578

67. Lu ZL, Dosher BA (1998) External noise distinguishes attention mechanisms. Vision Res 38: 1183–1198. PMID: 9666987

68. Pratte MS, Ling S, Swisher JD, Tong F (2013) How attention extracts objects from noise. J Neurophy-siol 110: 1346–1356. doi: 10.1152/jn.00127.2013 PMID: 23803331

69. Yan X-H, Magnasco MO (2012) Input-dependent wave attenuation in a critically-balanced model of cor-tex. PLoS One 7: e41419. doi: 10.1371/journal.pone.0041419 PMID: 22848489

D-Serine and Glycine Control of Neurotransmission in the Visual Cortex

PLOS ONE | DOI:10.1371/journal.pone.0151233 March 22, 2016 20 / 20